Author Topic: HV amplifier: putting a common-source (emitter) amplifier on opamp outputs  (Read 2026 times)

0 Members and 1 Guest are viewing this topic.

Offline WhalesTopic starter

  • Super Contributor
  • ***
  • Posts: 2121
  • Country: au
    • Halestrom
Every negative-feedback amplifier I have seen uses uses emitter followers (or source followers) on the outputs. Why is this so?

I think it might be easier to keep a follower stable in a feedback loop (because the transfer function between gate/base and output is more continuous and more spread out).

I also know that you can't make a BJT push/pull in a common emitter configuration: the two bases feed into each other.  You have to separate with previous stages OR use mosfets OR make it class A (one transistor) instead.

Context

I have a friend with a high-voltage piezo positioning device (~1uF claimed) that he wants to drive with 0-150V from DC to 10's of KHz.

At the moment I'm working on recreating a design that a previous team already claims works -- a big discrete HV opamp, all bipolar with a long tail pair on input and several stages of amplification to get to the output push/pull pair.  There are also several off-the-shelf products (mainly made by Apex) that would appear to meet many of our specs.

ie I know there's off the shelf stuff, this topic is for fun & education.

My whacky idea

Class A common-source amplifier on the output of a low-voltage opamp.  Performs the voltage shifting and heavy work for me:



I've made it mostly stable through the addition of the resistor between opamp inputs.  Source.

Seems very slew-rate limited, can't operate much above 10Khz.  This makes sense given the 100 ohm upper resistor + 1uF load.

You may notice the opamp inputs seem backwards.  This is because the common-source mosfet acts as an inverter.

Other physical notes:
  • High-wattage pullup resistor is probably going to be a collection of bulbs, kept under-powered so they don't change R too much.  Who cares about efficiency :D
  • I'm aware of SOA and have been reading the EEVBlog HV amp mega-topic.  The IRF640 might not be the best option here, it's just something my SPICE software has, and I plan to use several in parallel if necessary.
  • Source resistor is only approximate, it's being used to keep the sim happy.  End result will have some form of source and/or drain resistors (esp if I end up paralleling fets).

Simulation note: I'm using QUCS and it really doesn't like doing transient analysis on this circuit (gets stuck in infinite loops or complains of jacobians).  Several workarounds:
  • Source resistor (under mosfet): increasing this sometimes nukes the jacobians.
  • Small hinter circuitry (inside dotted box): I think this helped avoid infinite loops in analysis, not sure now.  I suspected that the solver was getting confused by the fact that changing the gate voltage on the mosfet didn't always change the voltage at the output, ie there was no gradient to descend.
  • I needed to set the 'minimum step size' in QUCS' transient simulation very high.  default: 1e-16, my setting: 1e-7.  This prevents the infinite loops (most of the time).
« Last Edit: December 16, 2019, 11:47:47 pm by Whales »
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
Re: HV amplifier: putting a common-source (emitter) amplifier on opamp outputs
« Reply #1 on: December 17, 2019, 12:46:36 am »
Some designs operate that way however the problem is that the common emitter/source stage adds voltage gain which depends both on operating point and the load impedance and this makes frequency compensation difficult.  The added voltage gain requires the open loop gain of the operational amplifier or gain-bandwidth product to be reduced; the resistor added between the inputs has that effect.

Two ways to improve the situation include:

1. Add emitter/source degeneration to better control the voltage gain.  This turns the transistor into a more linear voltage to current converter.
2. Add shunt feedback to the transistor stage to make its voltage gain fixed.  This is common with voltage boosted operational amplifier circuits.
 

Offline TurboTom

  • Super Contributor
  • ***
  • Posts: 1471
  • Country: de
Re: HV amplifier: putting a common-source (emitter) amplifier on opamp outputs
« Reply #2 on: December 17, 2019, 01:04:07 am »
In order to reduce power consumption, you may use a diode-dropper driven bipolar active pull-up. This will introduce some distortion upon cross-over (i.e. a slew rate change) but depending on the requirements, these problems usually are tolerable ones. I designed a scope clock with the high voltage deflection drivers arranged that way since I had to keep power consumption as low as possible. See here for the schematic to get an idea. The schematic shown settles within about 1µs after a 50V step. Quiescent current is less than 1mA for each channel.
« Last Edit: April 10, 2022, 12:51:51 pm by TurboTom »
 

Offline NiHaoMike

  • Super Contributor
  • ***
  • Posts: 9321
  • Country: us
  • "Don't turn it on - Take it apart!"
    • Facebook Page
Re: HV amplifier: putting a common-source (emitter) amplifier on opamp outputs
« Reply #3 on: December 17, 2019, 01:51:23 am »
A circuit similar to that (small flyback converter + discharge resistor switched in with a MOSFET) is used for driving piezoelectric expansion valves in some variable speed HVAC systems, but that only operates with a bandwidth of a few Hz at most. For something that has to operate into a few tens of kHz, the most obvious solution is to adapt an audio amplifier design.
Cryptocurrency has taught me to love math and at the same time be baffled by it.

Cryptocurrency lesson 0: Altcoins and Bitcoin are not the same thing.
 

Online Marco

  • Super Contributor
  • ***
  • Posts: 7043
  • Country: nl
Re: HV amplifier: putting a common-source (emitter) amplifier on opamp outputs
« Reply #4 on: December 17, 2019, 02:42:05 am »
Simulation note: I'm using QUCS and it really doesn't like doing transient analysis on this circuit

Try Simetrix, only downside is that more complex opamp models quickly take you over the node limit of the free version.
 

Offline duak

  • Super Contributor
  • ***
  • Posts: 1048
  • Country: ca
Re: HV amplifier: putting a common-source (emitter) amplifier on opamp outputs
« Reply #5 on: December 17, 2019, 03:42:11 am »
Whales, that's quite the industrial strength piezo actuator you've got there.  At 10 kHz and 150 V with a 1 uF cap the driver has to source then sink 1.5 A.

There are amplifiers that connect the output transistors in a common emitter arrangement.  One example is the quasi-complementary design that uses NPN-PNP transistors in the driver stage and all NPN (or PNP) output transistors. 

Dan Meyer designed a number of amplifiers (called Tigers) in the early 70's using a fully complementary output stage with a voltage gain of two.  Since the output stage had gain, the previous section didn't have to provide as great a voltage swing.  I liked this idea because, In theory, you can get a few more watts out for the same power supply.  I built a few but found they were more sensitive to output loading and could oscillate with capacitive loads more often than the standard emitter follower outputs.

« Last Edit: December 17, 2019, 03:44:20 am by duak »
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: HV amplifier: putting a common-source (emitter) amplifier on opamp outputs
« Reply #6 on: December 17, 2019, 04:02:54 am »
We can analyze this readily with analytical tools, no need for simulations. :-+

The op-amp is stable down to a certain noise gain.  At unity gain, it has whatever phase margin.  For the TL071, something like 60 degrees.  (Or if it's not a unity-gain-stable type, then some phase at whatever specified gain.)  If we introduce more gain to the loop, the margin goes away, and it at least peaks more, or oscillates outright.

And this is just for instantaneous (unlimited bandwidth) gain.  For real gain elements (like a transistor), there is some additional phase shift, which reduces phase margin even further.

The transistor has voltage gain.  With a grounded source (more or less), the voltage gain is close to R4 * gm (it's less by the output resistance, but that's generally quite high for FETs (channel length modulation factor) so we can ignore it).  With a source resistor R8, the maximum transconductance is 1/R8; however, the device's gm acts in series with R8, to realize the total equivalent.

The hybrid-pi model of the BJT uses r_e, a current-dependent resistor in the same position as R8, to represent the BJT's transconductance.  We do precisely the same here, except r_e * h_fe is not reflected back to the gate terminal as r_π, instead we only have gate capacitance and strays there.  Huh, I forget if they still call it a "hybrid-pi" model when it's a FET, but in any case, yeah, it works the same with these changes.

R8 usually dominates under class A bias conditions, because gm is usually large.  At ~0.7A as shown, gm looks to be around, oh, 0.25 S, which is comparable (larger, though not by much) to R8's conductance of 0.2 S.  The series total is 9 ohms or 0.11 S.

The voltage gain then is around 9, which isn't astonishing, and is actually more than matched by your feedback network.

That leaves the capacitance, which I would guess is dominant right around the opamp's cutoff, hence the hit to phase margin.

You can extend phase margin by strategically dulling the op-amp -- add an R+C from its output to -in, and adjust values until step response is ideal.  With a unity-gain-stable op-amp, this should give better performance than using R11.

The ideal values of the R+C will depend on capacitances as well as gain.  All of which are dependent here, so you can't compensate it very effectively over the full output range -- it will always be slower at low voltages (where capacitances are higher) and high voltages (where drain current is small).


Noise gain: the, well, gain of the op-amp's noise.  Typically, the feedback ratio.  Note that you've strapped a resistor across the inputs, which increases the noise gain without affecting the signal gain.  This does exactly what it sounds like it does -- noise goes up, including offset which is just DC "noise".  You've found this to be necessary, for precisely these reasons -- you need to reduce the amp's performance (unfortunately, this necessarily includes the noise floor too) to deal with the transistor gain in the loop.


General note -- collector- or drain-output amps aren't popular among amateurs, because they're harder to design and aren't really called for by their applications (mostly driving speakers, which are designed for very low source impedances, fractional ohms).  Every op-amp with rail-to-rail outputs, though, is such a type!

On that note, about output impedances -- the TL071 will itself have a follower style output, and an impedance around 100s of ohms around fT.  Consider increasing R7 to suit -- that way the op-amp has a reasonable fraction of its own output to read, without being completely swamped by gate capacitance.

And to improve gate drive speed, you can add a zero-offset follower if you like, or use a gruntier amp.  Using a newer transistor also helps, since MOSFET performance has improved by about 5x since the days of the IRF640 (though this comes at the price of smaller die area, so the power dissipation ratings tend to be lower for the same V,I ratings -- YMMV).

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
Re: HV amplifier: putting a common-source (emitter) amplifier on opamp outputs
« Reply #7 on: December 17, 2019, 04:25:00 am »
It may also be worth studying how voltage regulators intended to operate with low ESR output capacitors, like film and ceramic, make that work.

Even for an emitter/source follower output, a big ceramic or film capacitor is a difficult load.  With zero ESR, the output looks like an integrator which adds 90 degrees of phase lag at all frequencies causing instant oscillation.  The ESR of the capacitor adds a breakpoint in the integrator response providing stability.  For decoupling capacitors located at the load, the trace resistance provides the ESR.

The solution is to move the ESR to a point in series with the output before the capacitor.  Now high frequency AC feedback is taken before the series resistance and low frequency DC feedback is taken after it.  Integrated regulators can cheat here by providing the high frequency AC tap as part of the output transistor but it works just as well with discrete output stages with an added series resistor.
 

Offline WhalesTopic starter

  • Super Contributor
  • ***
  • Posts: 2121
  • Country: au
    • Halestrom
Re: HV amplifier: putting a common-source (emitter) amplifier on opamp outputs
« Reply #8 on: December 17, 2019, 05:03:35 am »
problem is that the common emitter/source stage adds voltage gain which depends both on operating point and the load impedance and this makes frequency compensation difficult.

I was aware of the 'depends on load' part: some playing around in sim revealed it to be much worse than I thought:



Imagine being the poor opamp that gets its output vs inputFB transfer function stolen from underneath it and replaced with a flat line.  Highway robbery.

"this makes frequency compensation difficult" -> thankyou, that's key.  Variable gain amplifier means I have to tune to suit worst case or at least very near to it.


2. Add shunt feedback to the transistor stage to make its voltage gain fixed.  This is common with voltage boosted operational amplifier circuits.

Hmm.  Am I reading this right when I think resistor between drain & gate (collector & base)?  I have always assumed that to be for biasing (for small AC signals), not for big trying to fix Vgain across large input & output swing ranges.


In order to reduce power consumption, you may use a diode-dropper driven bipolar active pull-up. [...] https://www.eevblog.com/forum/projects/maybe-risk-a-guess-what-thats-gonna-be/msg1810052/#msg1810052

Ooh, I like that.  Lazy top half and properly controlled bottom to make up for it.  Spiritually this makes me think of the H-bridge optimisation where you only PWM the lower transistors, but of course we're operating in a very different mode here.

Alas QUCS is resolutely shitting it's pants at that output stage and calling me all sorts of biblical names.  Changing to an ideal opamp with minimal gain hasn't helped, so I might try switching to some ideal FETs bypassed with 1M resistors instead.  I really don't like scrabbling in the dark, I'm not familiar with SPICE and I can't tell if it's my mind or the simulation that's the one actually diverging.

For something that has to operate into a few tens of kHz, the most obvious solution is to adapt an audio amplifier design.

Yep, that's pretty much the previous design I mention in the original post.  It's what I'm going to start with building, but I thought I'd have some fun contemplating some weirder/lower-partcount options.

Try Simetrix, only downside is that more complex opamp models quickly take you over the node limit of the free version.

Thanks for the suggestion. 

I am about to give Micro-Cap (free) a try.  It seems to have a very feature-filled GUI, an interesting mouse lazy-wirer and a reasonable number of parts models.

I have heard that ngspice has a better solver than QUCS (eg it has adaptive timesteps) so I might lookup common workflows for it.  It doesn't come with a GUI but it is in my repos.

I am open to all suggestions here, including "you're spicing it wrong".

Whales, that's quite the industrial strength piezo actuator you've got there.  At 10 kHz and 150 V with a 1 uF cap the driver has to source then sink 1.5 A.

Hahaha, not quite!  DC operation across the whole range + AC expected to be less than 10% of it.  Actuating mirrors in optics experiments, once roughtly positioned with DC the movement is expected to be less than 1 wavelength.

General note -- collector- or drain-output amps aren't popular among amateurs, [...] Every op-amp with rail-to-rail outputs, though, is such a type!

Ah, that makes complete sense.  It's hard/impossible to get within a few hundred mV of the rails without either going common emitter/source or exploiting low-current photon loopholes.


Quick edit:
On that note, about output impedances -- the TL071 will itself have a follower style output, and an impedance around 100s of ohms around fT.  Consider increasing R7 to suit -- that way the op-amp has a reasonable fraction of its own output to read, without being completely swamped by gate capacitance.

That explains why R7 (the gate series resistor) seemed to have no effect until I raised it nearer 1Kohm.  Thankyou, didn't think of the intrinsic impedence of the part.

I am not sure what you mean by "having a fraction of its own output to read" however.  I don't believe there's an internal feedback path, and if I increase R7 then the opamp output is going to slam/saturate either full +V or -V most of the time, which in my mind means its going to have to waste time slewing out of this condition.


@everything else  and everyone I have not replied to yet: thankyou, I'll get there, somewhat outnumbered :)
« Last Edit: December 17, 2019, 05:11:01 am by Whales »
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
Re: HV amplifier: putting a common-source (emitter) amplifier on opamp outputs
« Reply #9 on: December 17, 2019, 05:13:21 am »
2. Add shunt feedback to the transistor stage to make its voltage gain fixed.  This is common with voltage boosted operational amplifier circuits.

Hmm.  Am I reading this right when I think resistor between drain & gate (collector & base)?  I have always assumed that to be for biasing (ie to fix the Vgain for small AC signals), not for big swing ranges.

It changes the common emitter stage to an inverting voltage or current in and voltage out stage.  It is the transistor equivalent to an operational amplifier configured as an inverting amplifier.  Local feedback reduces the effect of the load impedance.

Shunt feedback in the output stage is common in "boosted" operational amplifier designs where a fixed gain discrete stage follows an operational amplifier.  Sometimes the feedback is only AC.
 

Offline Berni

  • Super Contributor
  • ***
  • Posts: 5050
  • Country: si
Re: HV amplifier: putting a common-source (emitter) amplifier on opamp outputs
« Reply #10 on: December 17, 2019, 06:33:53 am »
The easiest way of doing this is using a high voltage opamp like the LTC6090 to drive a complementary transistor follower stage to give the output the required oomph.

The datasheet for the LTC6090 even has a example schematic for that on page 20:
https://www.analog.com/media/en/technical-documentation/data-sheets/6090fe.pdf

You can probably also use a pair of BJT transistor in there too if you wanted, but for continuous operation you might need some very large transistors to have them survive all of the heat generated inside of them. Perhaps running multiple of these in parallel might be a good idea if the transistors get too hot. But for pulsed operation you can push a ton of power trough just a single transistor.

Indeed follower configurations are inherently more stable because they have a built in negative feedback loop. If the load on a follower suddenly disappears this causes the output to jerk upwards but in doing so also rapidly reduces the voltage on the base/gate and so the transistor shuts itself off without the opamp controlling it having to do a single thing. This makes such follower stages very stable, especially when driving weird loads that are very inductive, capacitive or non linear. But on the down side is that they don't provide any voltage gain so you need to provide them with a high voltage low current input signal. This can be done with a high voltage opamp like i suggested above, or using a low voltage opamp with an output transistor as you have shown, but use that transistor to only boost up the voltage, but let the follower stage after it actually drive the full output current.

Oh and physical layout of these sort of amplifiers can also be quite critical or you can accidentally end up with an oscillator instead.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf